> On April 21, 2012, 8:10 p.m., Gabe Black wrote:
> > The fp microops are pretty underdeveloped, but my impression was that there 
> > wasn't a lot of partial updating of the FP condition codes, and that would 
> > make this unnecessary. Could you please check how they're updated and 
> > report back? If setting the flags are optional but an all or nothing sort 
> > of thing, we don't need to add this code.

I think there are two fp microops that read/write the condition codes. Since 
the CC register
itself is not separate, any check that is introduced on that register, will be 
in force
even for fp microops. The check that I intend to introduce (it is there is one 
of the patches
posted on the board), works on the cc bits that would read/written by the 
microop. Hence, the need
for the change to the fp microops. I think it is possible to keep these changes 
to only those two
microops. I may try that out once we converge on the final solution for total 
set of patches.


- Nilay


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
http://reviews.gem5.org/r/1152/#review2570
-----------------------------------------------------------


On April 21, 2012, 1:15 p.m., Nilay Vaish wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> http://reviews.gem5.org/r/1152/
> -----------------------------------------------------------
> 
> (Updated April 21, 2012, 1:15 p.m.)
> 
> 
> Review request for Default.
> 
> 
> Description
> -------
> 
> Changeset 8960:5752699105b1
> ---------------------------
> X86: Add ext and cc fields to FP microops
> The cc and ext fields in integer microops hold the flags the microop is
> going to read and write respectively. These fields are being provided in
> FP microops. Later on this would help in deciding at run time whether or
> not certain flags need to be read or written. Currently these are set to 0.
> 
> 
> Diffs
> -----
> 
>   src/arch/x86/insts/microfpop.hh 0bba1c59b4d1 
>   src/arch/x86/isa/microops/fpop.isa 0bba1c59b4d1 
> 
> Diff: http://reviews.gem5.org/r/1152/diff/
> 
> 
> Testing
> -------
> 
> 
> Thanks,
> 
> Nilay Vaish
> 
>

_______________________________________________
gem5-dev mailing list
[email protected]
http://m5sim.org/mailman/listinfo/gem5-dev

Reply via email to